University of Virginia Library

Printing Methods and Textual Problems in A Midsummer Night's Dream Q1
by
Robert K. Turner, Jr.

I

A Midsummer Night's Dream Q1, printed "probably by Richard Bradock"[1] for Thomas Fisher in 1600, is a book of thirty-two unnumbered leaves. It collates A-H4. A1 is the title page, A1v is a blank; A2 bears the head title. The text, commencing on A2, is completed fourteen lines down on H4v, the remaining white space on that page being partially filled by a circular ornament. All leaves are signed with the conventional roman cap and arabic numeral, except F2 where an italic cap and H4 where a small cap are used. The speech-prefixes, which are indented, are set in italic caps and lower case, with some substitutions of small caps, and the stage-directions in italics with personae usually, but not always, in roman. Entrance directions are usually centered (exceptions occur on D2v, F1v, and H1v) and exit and other subordinate directions are placed in the margin unless they are fairly elaborate as they are on F4v and H1. There is nothing very striking about the typography; on cursory examination the book seems to be a run-of-the-mine Elizabethan dramatic quarto. Neither variations in spelling nor typographical abnornalities indicate that it was set up by more than one compositor.

A close look at the running-titles, however, gives us our first hint that the procedure adopted for printing was not so straightforward as may at first appear. Separately signed preliminaries, we have long known, were often printed after the text,[2] but when the preliminaries and the beginning of the text share sheet A we have some right to expect that sheet to have been set up and machined before the rest of the text. The running-titles, however, show us that sheet A of MND Q1 was the last sheet to go through the press. The book was worked in two skeleton-formes, one regularly imposing the inner and the other the outer formes.[3] The running-titles read on both recto and verso "A Midsommer nightes dreame." except on H3v where we find "A Midsommer nights dreame." Significant changes were made in two titles during the course of printing: (1) The "g" appearing in the title used on B3 and C3 (IV in the diagram below) was replaced at D4v, and the new type appears on E4v, F4v, G4v, H3, and A4v. (2) The title used on B4v and C4v (VII) is characterized by a broken "r" and a defective "e" in "dreame." At D3 a break in the "M" also appears, and the three defects are found together on E2v and F3 (where the two "e's" of "dreame" were exchanged in position). At G2v the "r" seems to have been replaced, and the "e" prints somewhat better than usual. When the title appears on H1, only the break in the "M" and the new "r" are evident, and only these two characteristics can be observed in the title as it appears on A3v.

It is then clear that sheet A was printed after sheet H, and with the aid of the following diagram, in which the running-titles are represented by roman numerals, we can understand why running-title IV, which is normally found in the same forme with VII and VIII, happened to be dissociated from them in sheet A:[4]

H(o), which, as I shall show below, must have been the first-printed forme of its sheet, evidently came from the press about the time the type pages of A(i) were ready for imposition, and it was thus convenient to impose A(i) in the H(o) skeleton. However, because A(i) contained a blank (A1v) and the head-title (A2), two of the H(o) running-titles, IV and V, were displaced. When the skeleton for A(o) was later made up, running-title IV, being available, was used together with two running-titles from the H(i) skeleton, that forme having been machined and its skeleton freed. Knowing that H(o) was sent to press before H(i) and that earlier in the book all the outer formes were imposed in the same skeleton used for H(o), and all the inner formes in the same one used for H(i), we can infer that outer formes of all sheets but A regularly preceded inner formes through the press.

An examination of the reappearances in various parts of the book of certain recognizable types shows that composition was by formes.[5] Let us first consider the implications of type reappearances in the first two sheets to be set. Type from B(o) is found in both formes of sheet C:

• d B1,18-C1v,8
• F B1,27-C1v,34
• k B2v,1-C1,8(?)
• b B2v,23-C4,26(?)
• f B2v,33-C3,33
• y B3,7-C2v, 7
• h B3,10-C1,1
• ſt B3,10-C2v,15
• B B3,sig.-C1v-23
• N B4v,7-C3,16
• ſſ B4v,24-C1v,5

whereas in C type from B(i) is found only in part of C(i):

• d B2,2-C3v,7
• f B2,14-C3v,16
• h B2,25-C3v,11
• y B2,27-C3v,34
• W B4,8-C4,7

Similarly, type from C(o) is found in both formes of D:

• d C1,7-D3,24
• g C1,9-D3,24
• d C1,28-D2v,31
• p C1,29-D4v,23
• y C2v,7-D4v,22
• y C2v,18-D3,17
• w C2v,24-D2v,30
• m C3,5-D2,14
• m C3,6-D2,14
• N C3,16-D1v,5
• u C4v,12-D1v,32
• k C4v,30-D1v,20
• u C4v,32-D1v,24

and type from C(i) in D only in D(i):

• ſſ C1v,5-D3v,23
• d C1v,7-D4,4
• n C1v,33-D4,28
• A C2,21-D3v,4
• n C2,24-D3v,3
• d C2,31-D4,10

When type reappears in this manner, composition cannot have been seriatim. Had it been so, B(o), to consider the forme which must have been distributed before the composition of sheet C had got very far along, could not have been made ready for the press until B4v had been set. The workman would then have started on sheet C, but we have evidence that B(o) had been worked off and distributed before he reached line 5 of C1v, that is, after he had set only a page of the new forme. Presumably B(i) would then have perfected its sheet, but this forme too was distributed during the setting of sheet C. The press, then, would have been delayed for at least the length of time required to set C3v and C4 before the first forme of C could have been imposed. We know, of course, that press delays sometimes occurred, but, if we examine the reappearances of sheet C type in sheet D, we will find that there too a delay would have resulted from seriatim setting and so through the rest of the book. It seems clear, then, that B(o) was completely set and sent to the press before B(i), that it was machined and ready for distribution before the composition of C(o) was begun, that B(i) was off the press and distributed during the setting of C(i) — in short, that the book generally was set by formes. But, as we shall see when we examine the type shortage evidence, this is not the whole story.

The same pattern of reappearances can be seen in most of the subsequent sheets. D(o) type is found in both formes of E:

• m D1,13-E3v,8(?)
• k D1,14-E4v,24
• k D1,21-E4v,23
• I D1,23-E4,10
• r D2v,12-E4,27
• W D2v,14-E1v,31
• k D2v,17-E1v,24
• m D4v,1-E2,13(?)

and D(i) type in E only in E(i):

• N D1v,5-E1v,35(?)
• k D1v,20-E2,8
• u D1v,24-E3v,23
• ſt D1v,25-E3v,31
• m D2,14-E3v,24
• y D2,27-E2,1
• B D2,32-E1v,7
• S D4,27-E4,30

E(o) type in both formes of F:

• d E1,5-F1v,27
• o E1,18-F2,14
• n E1,29-F2,26
• B E1,31-F1v,29
• d E2v,32-F2,10
• y E3,12-F4v,33
• m E3,17-F4v,23
• u E3,26-F4v,3
• b E3,29-F4v,21
• ſh E4v,25-F1v,7

and E(i) type in part of F(i):

• B E1v,7-F3v,26
• f E1v,13-F3v,16
• k E1v,24-F3v,25
• W E1v,31-F4,33

F(o) type in both formes of G:

• I F1,7-G1v,27
• M F1,13-G4v,9(?)
• ſh F1,18-G4v,10
• m F2v,5-G2,13(?)
• d F2v,28-G2,29
• b F3,6-G2,3
• W F3,7-G4v,8
• ſſ F3,16,G1v,29
• u F4v,3-G2,11
• M F4v,17-G1v,29

and F(i) type in G(i) [and also in H (o); see fn.11]:

• W F1v,2-G3v,3
• d F1v,27-G3v,5
• B F1v,29-G3v,21
• g F2,8-G3v,5
• d F2,10-G4,30

G(o) type, however, is found only in H(i) rather than in both formes of that sheet:

• u G1,5-H3v,8
• h G1,11-H1v,18
• ſt G1,12-H1v,32
• m G1,14-H2,3
• m G1,16-H2,1
• r G2v,19-H2,28
• ſi G2v,26-H2,15
• ſt G3,11-H4,14(?)
• m G4v,24-H4,22
• A G4v,26-H1v,13
• L G4v,27-H1v,31
• y G4v,31-H2,32

and G(i) type does not reappear before sheet A:

• y G1v,17-A2,13
• S G1v,26-A2v,5(?)
• B G1v,32-A2v,34(?)
• d G2,26-A3v,17
• w G3v,2-A2v,10(?)
• ſſ G4,2-A4,28(?)
• h G4,4-A2v,8(?)
• k G4,12-A3,26
• d G4,15-A3v,18
• ſ G4,22-A2v,10(?)
• d G4,30-A3,23(?)

H(o) type reappears in A(o), but I find no H(i) type in sheet A:

• k H1,1-A2v,17(?)
• f H1,5-A4v,8
• y H2v,17-A2v,20(?)
• y H2v,18-A2v,6
• h H2v,24-A3,21
• k H3,21-A2v,34
• L H3,31-A2v,5(?)

We can confirm, on this evidence, the inference drawn from our examination of the running-titles. It appears that B(o) was first composed and was machined while composition continued on B(i). The reappearance of B(o) type in both formes of C shows that B(o) was distributed before the composition of C(o) had gotten very far along; if we trust the identification of the "y" at C2v,7, we can say that it must have been distributed before that line was set. B(i), it is seen, was distributed after C(o) was completed but before the compositor had gotten more than half way through C(i).

This relationship between composition, presswork, and distribution seems to hold good throughout most of the book, but at sheet G certain abnormalities begin to appear: G(o), which should have been distributed before or early in the compostion of H (o), was apparently not distributed until the setting of that forme was completed, G(i) until after H(i) was composed,[6] and H(o) until after G(i) was completed. We thus have evidence that during the composition of sheet G the time relationship which had existed between the compositor and the press was disturbed, but we cannot without further examination tell what the cause of this disturbance was.

Of the evidence examined so far, two completely different interpretations

seem possible. The compositor may have finished the setting of the material cast off for G and H(o) so fast that he was able to impose H(o) before the press had worked off G(o). On the other hand, he may have been so delayed in setting that, although G(o) was returned from the press and available for distribution at the proper time, he deliberately put off its distribution until he could get a new forme, H(o), ready for imposition, thus avoiding or reducing a press delay. Sheet G does contain a little less material than other sheets; throughout the book the usual line count per page is thirty-five, and so it is in G except for G1, G1v, and G2v, which contain thirty-four lines and G2, which contains only thirty-two. The sheet as a whole contains six fewer lines than others, but this reduction, it seems to me, is more than balanced by prose passages on G1v, G2, and G2v. Thus we cannot account for an increase in the speed of composition on the grounds that the material in sheet G would have been easier to set than that in the earlier sheets.

To this point I have suggested that there were only two methods of composition—seriatim and by formes—available to the workman who set the type, and I have argued that the latter method was employed in MND Q1. Yet we must realize that when a compositor set a quarto by formes he did not necessarily have to set the type-pages in numerical sequence within the formes, although it is my distinct impression that this order was usually adopted. Nor did the compositor necessarily have to set either by formes or seriatim; he could, if he chose, combine the two methods. Although we can be reasonably sure that MND Q1 was in general set outer forme first, we might now see what we can tell about the order in which the pages of each forme were composed and whether the compositor ever deviated from strict adherence to the method which he generally followed.

In order to do this we must use type-shortage evidence in combination with the evidence of type reappearances. By itself the testimony of shortages is, I believe, less reliable than that of any other bibliographical technique. We generally assume that when a compositor ran short of type of a certain kind, say roman capital A, he substituted for it another appropriate kind of type, say italic capital A or small capital a, and that he continued his deliberate use of wrong-font letters until his supply of proper types was replenished by distribution. There is no doubt that compositors substituted in this manner: we can in some instances determine exactly how many types of a particular kind were in the case; we can see these types being used up, substitutions for them

being made, the original supply being reused after distribution, and then the cycle repeating itself.[7]

What sometimes complicates matters is that compositors apparently did not always wait until the supply of regular type was completely exhausted before they began to substitute, and then when distributing they did not always separate the substituted from the regular type but evidently distributed both into the same box in such a way that the two sorts later appear in a more-or-less random mixture. Even when attempts seem to have been made to keep the two sorts separated, once substituting began compositors seem occasionally to have substituted when they had plenty of the regular type on hand, probably because the supply of regular type became fouled during distribution. However, the reliability of type shortage evidence can be increased when we evaluate it in the light of type reappearances, but even here we can be forced away from the most desirable position by occasionally having to take into account the evidence of only one or two reappearing types and sometimes having to argue from the non-reappearance of type. Both are bad policies because mistakes in individual type identifications are easy to make and reappearances are easy to miss.

With these precautions in mind, let us examine the type substitution in MND Q1. In sheet B we find that a shortage of roman capital A was made good by the substitution of small capital A, as follows (here and subsequently, numbers to the left of the stroke represent types of the proper kind and those to the right substituted types):[8]

     

		B(o)				B(i)
	1	2v	3	4v	1v	2	3v	4
A/a	5/0	2/0	5/0	10/0	6/0	8/1	8/5	4/1

Here it seems evident that in order to stretch a dwindling supply of A's the workman began to substitute occasionally on B2 and continued to do so through the rest of the sheet. The pattern is consistent with our earlier conclusion that B(o) was first completed, and we can say that B2, B3v, and B4 of B(i) were probably composed after the rest of the type-pages in the sheet. It is a safe guess that all of B(o) was set before work began on B(i), but we cannot absolutely rule out such an order as B1-B1v-B2v-B3-B4v-B2, etc.

If this seems a borrowing of trouble, we might look at the next sheet, where once again the A's run short:

     

		C(o)				C(i)
	1	2v	3	4v	1v	2	3v	4
A/a	10/0	0/4	0/2	0/4	9/0	6/0	4/0	2/0

Type reappearances tell us that B(o) was almost certainly distributed by the time C2v was set and possibly before much of C1 was set (on the dubious evidence of the "k" at C1,8). The type substitution pattern indicates that the earlier point is the correct one. But B(o) contained twenty-two A's. Why, then, did the compositor begin to substitute on C2v when he should have had twelve A's available even if he had completely exhausted his supply in completing B(i)? Evidently he did not set in the sequence indicated above, but as follows:      

				C
	1	1v	2	2v	3	4v	3v	4
A/a	10/0	9/0	6/0	0/4	0/2	0/4	4/0	2/0

Here twenty-five A's are used before substitution begins at C2v, a reasonable number if twenty-two came from B(o) and a few remained after B(i) was completed. Moreover, we found no B(i) types in any pages of C(o) or on C1v or C2. Apparently the compositor's intention was to set the sheet seriatim, but, upon completing C3 and realizing that he could not get the inner forme completed in time to avoid a press delay, he set C4v, thus completing the outer forme first. The supply of A's was replenished by the distribution of B(i) after C4v was set and before work started on C3v.

The evidence in sheet D is weaker, but the best indications are that the compositor did not try to revert to seriatim setting:

     

		D(o)			D(i)
	1	2v	3	4v	1v	2	3v	4
A/a	7/0	5/0	8/0	5/0	--	6/1	5/0	7/0

Twenty-six A's were returned to the case with the distribution of B(i); six were required for C3v and C4 and seven for D1, at which point, according to the type reappearances, C(o), containing ten A's, was distributed. There were, then, at least twenty-three A's in the case when work started on D2v. By the time the single A was introduced on the next-to-last line of D2, twenty-four A's had been used, a close enough tally. As type reappearances indicate, C(i) was distributed at this point. We can thus be reasonably sure that, since D2, D3v, and D4 were set after the other pages of the sheet, the order shown is correct.

In sheet E two new shortages are found:

         

			E(o)			E(i)
	1	2v	3	4v	1v	2	3v	4
A/a	7/6	0/2	3/2	4/0	3/0	5/0	2/0	6/0
H/h	1/1	2/1	0/3	1/1	3/0	1/0	1/0	2/3
H/h	3/0	9/0	4/0	11/0	4/0	7/1	3/6	0/7

Type reappearances tell us that D(o) was distributed before or during the setting of E4v: as we have noted above, one D(o) type appears at E4v,23 and another at E4v,24. Since the first A on that page occurs at line 14 and since the small cap H for roman capital H occurs at line 11 and the one roman capital H (used incorrectly for an H) at line 24, we can probably believe that the compositor stopped work temporarily at line 12 or 13 to distribute D(o). The number of types used is close to the number that we can estimate to have been available. The distribution of C(i) had returned twenty-one A's and nine H's to the case, and we have reason to think that the supply of A's, at least, was low at the time of distribution. Nineteen A's were used on D3v, D4, and E1, where the substitution commences, and three more on E3, just before the next distribution. Seven H's were consumed in D3v and D4 and four more used before the supply was replenished.

Type reappearances also indicate that D(i) was distributed very shortly after D(o), the first type from this forme reappearing on E1v. Before the composition of E started, the case was resupplied with eighteen A's, enough to finish out the forme with no substitutions, but only eight H's and no H's. Thus small-cap h's were introduced in place of both these letters late in the forme. Once again it looks as though composition must have gone in normal sequential order within the formes.

The situation in sheet F is not quite so clear. Here another series of substitution begins:

           

		F(o)			F(i)
	1	2v	3	4v	1v	2	3v	4
A/a	7/0	5/0	3/0	2/0	8/0	4/0	5/1	0/5
H/h	1/0	3/0	2/0	1/0	2/0	1/2	2/0	1/0
H/h	0/4	0/1	--	4/1	--	1/0	--	0/1
T/t	--	--	4/0	6/1	--	--	2/1	3/4

Type reappearances tell us that E(o) was distributed between the setting of F3 and F4v, testimony that is confirmed by the reappearance of H's on that page. The substitution for T after F3 is to be expected because E(o) had contained but one piece of this type. Yet, twenty-seven H's would have been returned from E(o) and there should have been no substitution for this letter on F4v or F4. The h on F4v, however, is found in line 1 and may have been set before the distribution was made, and that on F4 is found in the catchword, a fact which may somehow make a difference. These two aberrant substitutions we may be able to accept, but the presence of H's on F1, F2v, and F3 also creates a problem, since substitutions for this letter had begun on E4v and the resumption of the exclusive use of H suggests that the distribution may have occurred at F1. The whole affair is rather unsatisfactory, but I would rather believe that the compositor, perhaps influenced by the genuine shortage of H's, unnecessarily set a few h's for H on E4 (he had, after all, at least thirteen H's in the case when he began E (i)—six from D(o) and eight from D(i) less one used on E4v) than that, if E(o) had been distributed before F1, he would continue to substitute unnecessarily for H through two pages and that no recognizable types from E(o) could be found until F4v. I believe, then, that the order shown for F(o) is probably correct.[9]

Upon the completion of this forme, the workman seems to have gone ahead with F(i), setting F1v, F2, and perhaps part of F3v, and then suspending composition to distribute E1v, as type reappearances suggest. From this single type page only three A's, three H's, and no T's were returned to the case, thus necessitating substitution for the A's and T's required by F3v and F4. The rest of the standing type seems to have been distributed after F(i) was imposed.[10]

Sheet G too seems to have been set in regular order within the formes:

           

		G(o)			G(i)
	1	2v	3	4v	1v	2	3v	4
A/a	4/0	5/0	4/0	8/0	7/0	1/0	7/0	4/2
H/h	2/0	1/0	2/4	1/3	1/0	3/0	2/0	1/0
H/h	4/1	1/1	1/1	1/0	6/0	--	--	2/0
T/t	--	2/1	1/4	3/4	--	4/0	3/3	0/3

When E2, E3v, and E4 were distributed just before the setting of G1 began, the supply of A's, H's, and T's seems to have been nearly exhausted. The distribution returned to the case thirteen A's, four H's, ten H's, and two T's. Type reappearances indicate that at least two pages of F(o)—F1 and F3—were distributed between the setting of G3 and G4v. In the composition of the first three pages of G(o), exactly thirteen A's were used, but the newly provided H's and T's were inadequate and substitutions were required for both letters. The occasional use of h's for H's throughout G1, G2v, and G3 was unnecessary since there were about thirty-six of the italic pieces in the case when the composition of G(o) began, and I can account for their appearance only on the supposition that some small caps had accidentally been mixed with the italic. Because the distribution of F1 and F3 provided ten A's but only three H's and four T's, we find no substitution on F4v in the A's but continued substitution in the H's and T's. The distribution of the other two pages of F(o) apparently took place after G(o) was imposed and brought in seven A's, four H's, four H's, and six T's, which type supplied the demands of the pages of G(i) until, as type reappearances suggest, F1v and F2 were distributed between the setting of G2 and G3v. These two pages of F(i) returned to the case twelve A's, of which eleven were consumed before a's were introduced near the bottom of G4; three H's and one H, enough to hold good through the rest of the sheet; but no T's, thus causing substitution for this letter to begin on G3v. The remaining pages of F(i) were distributed after the imposition of G(i).[11]

In the two remaining sheets, H and A, I can see no indication that the type pages were not set in sequential order within the formes, although the type-shortage evidence is not so clear as it might be. It would seem that by this time all of the regular boxes were becoming fouled with the small caps and that substitutions were sometimes made accidentally. In sheet H a new shortage appears:

             

		H(o)				H(i)
	1	2v	3	4v	1v	2	3v	4
A/a	4/0	3/0	2/6	1/2	5/0	4/0	4/0	5/0
H/h	2/0	--	2/1	--	1/0	5/0	--	1/2
H/h	--	--	--	--	1/0	1/0	1/0	--
T/t	7/1	0/2	0/4	--	1/8	1/1	1/0	0/1
P/p	3/0	4/0	1/2	--	2/6	--	2/0	--

             

		A(i)				A(o)
	1v	2	3v	4	1	2v	3	4v
A/a	--	6/0	5/3	2/3	--	5/1	1/0	4/0
H/h	--	4/0	--	1/1	--	--	--	1/0
H/h	--	2/1	6/0	6/0	--	2/2	4/0	6/2
T/t	--	1/2	0/1	--	1/0	1/1	3/0	1/0
P/p	--	1/0	--	--	--	--	--	--

Type reappearances testify that F3v and F4 (containing five A's, three H's, no H's, five T's, and one P) were distributed at H1, G(o) (twenty-one A's, six H's, seven H's, six T's, and six P's) at H1v, G(i) (nineteen A's, seven H's, eight H's, seven T's, and fourteen P's) at A2, and H(o) (ten A's, four H's, no H's, and seven T's) at A1 or A2v. The substitutions throughout H(o) bear out the time of distribution of G(o), but after this the substitutions seem to occur rather erratically. It is possible that some pages within the formes may have been set out of order (e.g., A3-A4v-A2v-A1), but I can see no evidence strong enough to determine the matter. The occasional substitution in the text of t for T on H3, H4v, H2, H3v, H4, and A3 suggests that these pages may have been the last set within their respective formes.

With the information we have thus gained, we can with some confidence chart the progress of composition as follows:

From this we can draw several inferences. It seems likely that the compositor, working on the assumption that composition and presswork could stay more-or-less in balance, originally intended to follow the conventional procedure for setting by formes—to compose two formes, distribute the first, set the third, distribute the second, and so on. That B(o) was machined in the time required to set the four type pages of B(i) or perhaps even a little less time is indicated by its distribution before the composition of C began. Thus the speed of the press, which barring accidents would have remained fairly constant, is established as the rate at which about four type pages could be composed. The compositor could alter this time relationship in his own favor by increasing his normal rate of work or by setting quick pages—pages with short lines or plenty of white space—but I doubt that in setting normal material he could gain more than a page in a forme, if that much. His attempt to shift to seriatim setting in sheet C caused him to delay the distribution of B(i) until he could get B(o) ready for the press, for that forme was probably returned to him while he was setting C2v or C3. The distribution of C(o) only three type-pages after B(i) is a further indication that the press was working at something like its original rate and that the compositor deliberately delayed the distribution of B(i).

Thus we have reason to believe that throughout the book certain formes may have been distributed later than they should have been, not because the compositor had gained on the press, but because he wanted to get a new forme imposed or get within a page or so of imposition before he temporarily stopped composing to distribute. Further, we can understand the nearly simultaneous distribution of the two formes of D; the piecemeal distribution of E(i), F(o), and F(i), attempts to get some type on hand during the composition of a forme without using any more time than necessary; and the distribution of G(o), H(o), and G(i) four type pages late in each instance.

II

Some years ago Professor J. Dover Wilson edited the New Cambridge MND on the theory that an original of 1592 was revised in 1594 and again in 1598, perhaps then for the marriage of the Earl of Southampton to Elizabeth Vernon.[12] The copy for Q1, he believed, was the prompt book, which was written in Shakespeare's autograph and which contained manuscript pages from all three periods of composition.

Wilson's theory has not won acceptance in its entirety; Sir Walter Greg, for example, found difficulty with Wilson's notion of multi-level revisions in which there are perceptible (and dateable) stylistic differences and suggested that the Q1 copy was probably foul papers.[13] If so, however, there could easily have been revised passages present, even though they may not have been composed in accordance with Wilson's scheme. As Greg said, Greg does not comment on Wilson's contention that other parts of the text were also revised or added to, but if the copy was foul papers with occasional marginal additions, we could explain the compositor's difficulties as resulting from his being slowed down by patches of difficult material. If the quarto is in part mislined, there is reason to think that the copy may have been similarly mislined, and an important part of Wilson's argument for revision is based on inference from quarto back to MS mislineation. Yet, as the quarto was almost completely set by formes, the compositor may sometimes have tampered with the MS lineation, deliberately compressing or expanding the text in order to make it conform with the space limits established by casting-off. Since we have some idea of how each sheet was set, however, we should be able to tell to what extent the compositor was likely to alter MS lineation as he dealt with mechanical problems created by his method of composing.

In B(o), the first forme composed, there are two minor instances of mislineation. The first occurs in the first eight lines of the fairy's speech (B3,14-17;II.i.2-9), which appear in the quarto as follows (correct verse line endings are indicated by a stroke):

The second is found on the last page of the forme (B4v,20-21; II.i.115-116):

About the first of these one cannot be positive, but it looks as though the lines probably stood in the MS as they do in the quarto, the quatrain being lined as a couplet and the couplets as single lines of verse. If this is true, the couplets of which the rest of the speech is composed, so lined in the quarto, probably were written in single lines in the MS, but were divided by the compositor who, having had to turn over the last word of B3,17, decided that the MS lines were going to be too long for his measure. The compositor could, of course, do about as he chose with this material since he was working on the first forme of the sheet. In the second case, the mislineation may have resulted from the compositor's carelessness; but later in the play (at F3,26; G1,22; G3v,10; and G3v,18) we find other examples of very much the same thing, which just possibly suggests that the MS rather than the workman was at fault.

When B(o) was completed, the limits of B1v,2 and B3v,4 had been established; thus we have more reason to think that the compositor might juggle the text of the inner forme to make it get in these limits. Almost certainly he was doing just this when he set a short speech of Bottom's and one of Peter Quince's in a single line of type at the foot of B2 (B2,33; I.ii.52-53). But it appears that Bottom's Ercles speech (B2,15-18; I.ii.27-34) which is printed as prose was probably written as prose in the copy, since no provision was made in the casting-off for it to be set as verse. One further mislineation occurs in B(i): in the middle of B3v (l. 16; II.i.42-43), we find the beginning of one of Puck's couplet speeches in this arrangement:

As the speech goes on for fifteen correct lines, the compositor must have planned to set it in couplets and allowed space for them when he established the place in the text at which to begin B4v even though, by analogy with the fairy's speech on B3, the couplets were written in single lines in the MS. In this instance, however, he apparently thought that he could squeeze the first complete verse line into the same line of type with the half-line of verse which begins the speech, a calculation which, as the turn-over shows, was none too accurate.

Since, as we have seen, the compositor seems to have had to adjust the MS lineation in sheet B, we can better understand his decision to shift to seriatim setting in sheet C even though in doing so he penalized himself in his time relationship with the press. C4v was set out of seriatim order, but no serious problems seem to have arisen. The only instance of mislineation is found at C1v,12 (II.i.175-176) where the compositor chose to set a line and a half of verse in one line of type. It is in this part of the text, however, that Wilson thinks heavy revision to have been made—II.i.1-147 belonging to one level of composition, ll. 148-187 to a second, ll. 188-246 to a third, and ll. 247-268 (end of scene) and II.ii.1-42 to the second again. As C1 begins at II.i.130 and C3 ends at II.ii.38, the seriatim setting of the first five pages of the sheet may indeed suggest that the workman was confronted here with a particularly nightmarish piece of composite copy.

In D(o) we find two more minor instances of mislining—a prose speech of Peter Quince's at D2v,8-9 (III.i.93-94) set in part as verse, doubtless under the influence of Thisbe's immediately preceding verse speech, and at D4v,17 (III.ii.48-49) a line and a half of verse set in one line of type. The textual material included in this sheet runs as follows:

       

D1	II.ii.141-III.i.11	D1v	III.i.11-III.i.47
D2v	III.i.85-III.i.120	D2	III.i.48-III.i.84
D3	III.i.121-III.i.153	D3v	III.i.154-III.i.190
D4v	III.ii.33-III.ii.67	D4	III.i.191-III.ii.32

It is interesting that all the material in D(i), the later forme, is perfectly lined, although it is set solid and includes on D1v and D2 much more prose than verse. According to Wilson, all of III.i and the first forty lines of III.ii belong to the same stratum of composition and underwent small, if any, revision.[15] The exact fitting of the material in the quarto suggests that the manuscript at this point was regular enough to permit quite accurate casting-off. Wilson also believes the half-line at D4v,17 to mark an abridgement, but this is a matter upon which our analysis provides no additional information.

However, the piecemeal distribution of E(i), beginning with E1v after the setting of F2 or part of F3v, suggests that work on F(o) and the first two pages of F(i) went slowly; and this is another part of the

text (III.ii, ending near the foot of F2v) which Wilson thinks to have been considerably worked over. He sees the irregular lining at E1,13-14 (III.ii.80-81), a shift from couplets to blank verse at E1v,27 (III.ii.127), and a short line at E3v,19 (III.ii.256) as indications of abridgement, addition, and cancellation. The first is the only mislined passage in the sheet, and, since it occurs on the first page of the forme and does not affect the total number of lines in the page, it seems virtually impossible for the passage to have been deliberately mislined for mechanical reasons. Nor does there seem to be any likelihood that the two subsequent aberrations in the text, even though they are found in the inner forme, arose from the compositor's tampering. At F1v,31-32 (III.ii.396-399) we find Puck's four-line speech set in two lines of type, an indication, according to Wilson (pp. 125,130) that they were squeezed in at the foot of a page of revised MS. As two lines only must have been allowed for this speech in the casting-off, the rest of F1v and F2 being set solid, they were doubtless written in two lines in the manuscript, but it seems much less certain that they were written so as to crowd them on an MS page when we remember that the fairy's speech on B3, which is in the same meter, seems to have been similarly lined in the MS.

IV.i begins near the bottom of F2v and ends seven lines down on G2. About this scene Wilson says in part:

The first of these speeches is at F3, 18-19; it appears as follows: The second is at F3, 26-27: Because he did not turn "loue" over, his usual practice when a verse line was too long for the measure, the compositor probably thought the first of these speeches to be prose. We have seen him at D2v,8-9 line prose as verse, and it does not seem unlikely that here, under the influence of Bottom's immediately preceding prose speech, he would have mistaken verse for prose, especially since the line was probably written in one line in the MS. In the second case, however, the turning-over of "hoord" and the capitalization of "And" make it clear that the compositor knew he was dealing with verse, although he handled it incorrectly. We have here a case very like that previously observed at B4v,20, only in this instance I think it would have been very difficult for the compositor to have lined in this manner as the result of carelessness. The lines probably stood in the MS as they do in the quarto, but neither this mislineation nor that at F3,18 provides much evidence for the recopying of an earlier draft. However, the material in F3v and F4 (IV.i.44-IV.i.110), all verse, is perfectly lined and set solid, an indication that the MS, recopied or not, was regular enough for an accurate line count.

Sheet G includes the following textual material:

       

G1	IV.i.144-IV.i.177	G1v	IV.i.178-IV.i.212
G2v	IV.ii.22-V.i.15	G2	IV.i.212-IV.ii.22
G3	V.i.15-V.i.48	G3v	V.i.49-V.i.83
G4v	V.i.118-V.i.150	G4	V.i.84-V.i.117

In this section of the text the most serious mislining occurs, and, as we have noted earlier, on this evidence Wilson argues most strongly for revision in the copy which underlay the quarto. From the point of view of printing mechanics, the sheet has another feature (in addition to the mislining) which is distinctly odd: the number of lines of type per page, earlier a consistent thirty-five, was reduced to thirty-four on G1, G1v, and G2v and to thirty-two on G2. Because the mislining is present on G3v (which contains thirty-five lines of type), it is clear that there is no absolute relationship between the incorrect lineation and the reduction of the number of lines of type per page in the first four pages of the sheet—that is, the lines of type per page were not reduced simply because the compositor compressed material that would have occupied more space if it had been lined correctly.

Let us look more closely at the passages in question. At G1,21-23 (IV.i.164-166) we find:

Here, as at B4v,20 and F3,26, the compositor had nothing to gain by altering the lineation since his version occupies just the same number of lines as the correct version, nor does it seem very likely that his error resulted from a failure of memory after he had seen correct lineation in the MS. Wilson suggests (p. 135) that the lines were written in the margin of the manuscript "to take the place of a longer cancelled passage." Certainly the indications are that the quarto lining reflects the lining of the copy.

On G2v and G3, where there are two lengthier mislined passages, the effect of the mislining is different. The first is found at G2v,26-G3,3 (V.i.5-18); asterisks indicate lines which Wilson (pp. 80-81) thinks to have belonged to the earlier layer of composition:

The second is at G3,16-24 (V.i.29-38): In both instances the text is compressed, a total of twenty-five lines of verse (according to the quarto's standard of lineation) being set in twenty-one lines of type. Neverthless, as was true on G1, there seems to be no mechanical reason here for the mislineation, since the compositor was setting the first forme. Once more it seems likely that the lineation of the quarto reflects the lineation of the copy: there is no mechanical reason why the lines should not have been, as Wilson believes (p. 85), "written on the margin of the MS, in such space as could be found."

In G(i) there are further instances of mislineation. At G1v,7 (IV.i. 184-185) a short verse line is set in the same line of type with a complete verse line, in just the way we have seen earlier at B3v,16 and C1v,12. There is no evidence to tell us positively whether the copy stood this way—the compositor may either have been setting what he saw or hedging against faulty casting off—but the occurrence of the same kind of mislineation on C1v, which was set seriatim, creates a presumption that the copy was so lined. Similarly, we find at G1v,22-23 (IV.i.197-198) that there are two lines of verse set as prose (cf. F3,18-19), probably under the influence of Bottom's immediately following prose speech; no space is saved by the mislineation. G3v contains four mislined passages, the first at G3v,10-12 (V.i.58-60):

the second at G3v,18-22 (V.i.66-70): the third at G3v,29-30 (V.i.76-78): and the fourth at G3v,34-35 (V.i.81-83): Since we have seen this sort of mislining in the outer forme, we are probably safe in thinking that here too the mislineation arose from the copy and not from the compositor's efforts to compress his material into a fixed space. Had the workman been making adjustments for purely mechanical reasons, he probably would have done so on G4 as well as G3v, but the text on G4 is correctly lined.

Although detailed reconstruction of a compositor's procedures must be in part speculative, I believe we can see about what happened if we try to follow our man step-by-step through his work on the sheet. We know that he set G(o) first and that he was probably having to work as quickly as he could to keep up with the press. Before beginning G1 he

must have cast off the whole sheet, counting out thirty-five lines for each page and marking page-endings in the copy. Had he been under less pressure, he probably could have worked out the correct lineation of the mislined passages (we saw him making correct adjustments in the lineation of his copy in sheet B) which, as Wilson has noticed, were probably marginal additions; but as it was he counted them as they were lined in the MS. If this was his procedure, however, he evidently counted in some material that he later did not set, and it does not seem unlikely, if he was working hastily, that he may have failed to notice that some lines here and there were supposed to be cancelled. Wilson detects abridgements in the text at G1, 22-23 (IV.i.165-166) in the outer forme and at G1v,22-23 (IV.i.197-198) in the inner. It seems quite possible, then, that the compositor discovered during the course of setting G1 that he had in his casting-off included a cancelled line. He thus reduced the number of lines of type on that page by one, preferring to set the page short than to readjust the subsequent pageendings already marked. I take it that the lack of one line of type on G2v arises from similar causes, although there the miscalculation probably sprang from a failure to count out the marginal material correctly; and that on G1v a cancelled passage of about four lines was provided for in the estimate of space but was not set up, thus requiring the omission of one line of type on G1v and the opening up of the material on G2. An incomplete prose line at the foot of G2 (l. 31) indicates that the workman had to strain a little to marry that page with G2v, which had already been composed.

The instances of mislineation in sheets H and A (with one exception) are so like those we have already examined that we may summarize them:

• H1,33-H1v,1 (V.i.182-185): Prose mistaken for verse.
• H1v,7 (V.i.193): Two lines of verse in one line of type; probably MS lineation.
• H2,35 (V.i.262): Two lines of verse in one line of type; probably compositor's lineation (cf. B2,33).
• H4,7 (V.i.394): Couplet set in one line of type; probably MS lineation.
• A2v,33) (I.i.53): Two short lines in one line of type; responsibility doubtful.

At A2v,2 and 5 (I.i.24 and 26) Stand forth Demetrius and Stand forth Lisander are centered rather than printed as part of the text as meter requires. Since he set the proper names in italics rather than in roman, the compositor evidently did not take these phrases as stage directions, but lined them about as they were lined in the copy. Wilson (p. 105) notes that "they were probably written as separate half-lines in the original MS, to denote deliberate utterance . . ." or, we may add, to cue obvious stage business.

We have seen that the quarto was set largely by formes under adverse conditions probably caused by sections of difficult copy. In only two places (B2,33 and H2,35) does there seem to be any likelihood that the compositor juggled the lineation of the text in order to fit copy to a predetermined space.[17] The bibliographical evidence seems to me to point toward the kind of heavily revised manuscript described by Professor Wilson as copy for the quarto, although it does not, of course, in itself lend any support to his distinction of different levels of style in the revisions. There is, as Sir Walter Greg has noted, no indication that the manuscript ever served as a prompt book, and the presence of revision may lead one to think of late-stage foul papers as copy.

notes